Biomedical Engineering Reference
In-Depth Information
Target window
Baseline window
f
j
1
f
j
(, )
j
ft
(, )
j
ft
k
c
f
j
1
t
t
t
t
k
k
1
k
2
c
time
Fig. 3.1
Depiction of time-frequency domain discretization. The target window is set at the one
represented by
(
f
j
,
t
k
)
, and the baseline window is set at the one represented by
(
f
j
,
t
c
)
3.7.4 Five-Dimensional Brain Imaging
Using the narrow-band dual-state beamformer, we can implement the five-
dimensional (time-frequency-space) imaging of brain activities [
7
]. In this imple-
mentation, we use the sliding window method in which the target window is moved
along the time and frequency directions. The discretization of the time-frequency
domain is depicted in Fig.
3.1
.
We assign the window denoted
(
f
j
,
t
k
)
to the target time-frequency window, and
the window denoted
to the baseline window. By implementing the narrow-
band dual-state beamformer using these two time-frequency windows, we obtain
the pseudo F-ratio image,
F
(
f
j
,
t
c
)
, which represents the source power difference
at the frequency
f
j
between the time windows at
t
k
and
t
c
. If we move the target
window along the time and frequency directions, we can obtain the five-dimensional
source power difference map of the induced brain activity,
F
(
r
,
f
j
,
t
k
)
(
r
,
f
j
,
t
k
)
, where
j
=
1
K
, where
N
f
is the number of frequency bins and
K
is
the number of time windows.
,...,
N
f
and
k
=
1
,...,
3.8 Nonadaptive Spatial Filters
3.8.1 Minimum-Norm Filter
There is a different class of beamformers that uses the nonadaptive weight, the
weight computed only using the sensor lead field. These are customarily called as
